Many national policies and laws, such as reducing health disparities and civil rights enforcement, rely on the measurement of the population by race and ethnicity. There are, however, growing doubts about the reliability and accuracy of race and ethnic data in American social statistics. The proposed research will address some of the major underlying issues that influence the conceptualization and measurement of race and ethnic identities in the United States with data from the 2000 Census, the National Longitudinal Survey of Adolescent Health (Add Health) and the Beyond High School Survey in Washington State. The specific aims of this research project are to: (1) develop a typology of single and multiracial identities that can be systematically compared across different survey questions, administrative records, and perceived race/ethnic classifications, (2) explore variations in the patterns of response to different survey questions on race and ethnic identities, including an analysis of which persons change their identities in response to different questions, (3) estimate a baseline model of individual race/ethnic identity(ies) as a function of parental ancestry and appearance (based on photographs) and to investigate how the parameters of this model vary by family background characteristics (birthplace, home language, socioeconomic status), (4) test hypotheses of how context, reciprocity, and peer identities influence individual ethnic choices, (5) estimate patterns of consistency and change in race and ethnic identities over historical time and the early life course, (6) estimate the impact of different definitions of race/ethnic boundaries (e.g., single vs. multiple identities) on race and ethnic disparities in educational aspirations and attainment, and (7) develop a theoretical schema of the underlying principles that influence race and ethnic identities in early twenty-first century United States.